Introduction
Soybean (Glycine max (L.) Merr.) is the most important grain legume crop in the world. Cultivated soybean seeds contain ~40% protein and ~20% lipid, and its crop economic value is dependent on the concentration of protein and oil (Brumm and Hurburgh, Reference Brumm and Hurburgh1990). The relationship between the protein, oil and sugar components of mature soybean seed has been the subject of numerous investigations (Hurburgh et al., Reference Hurburgh, Brumm, Guinn and Hartwig1990; Hurburgh, Reference Hurburgh1994; Piper and Boote, Reference Piper and Boote1999; Yaklich et al., Reference Yaklich, Vinyard, Camp and Douglass2002; Brumm and Hurburgh, Reference Brumm and Hurburgh2006; Dardanelli et al., Reference Dardanelli, Balzarini, Martinez, Cuniberti, Resnik, Ramunda, Herrero and Baigorri2006). Endemic over a wide range of East Asia, wild soybean (Glycine soja Sieb. and Zucc.) is the closest relative of cultivated soybeans (G. max) but both have prominent morphological and physiological differences, known as the domestication syndrome (Broich and Palmer, Reference Broich and Palmer1980; Hyten et al., Reference Hyten, Song, Zhu, Choi, Nelson and Costa2006). However, the significant loss of diversity among high-yielding adapted lines ultimately inhibits future genetic gains in productivity, broadens susceptibility to new pests and diseases, and acts as a threat to food security (Carter et al., Reference Carter, Nelson, Sneller, Cui, Boerma and Specht2004; Hyten et al., Reference Hyten, Song, Zhu, Choi, Nelson and Costa2006). The narrowed genetic base in cultivated soybean limits improvements in yield and quality (Hajjar and Hodgkin, Reference Hajjar and Hodgkin2007; Chung and Singh, Reference Chung and Singh2008). In contrast to modern soybean cultivars, wild soybeans are genetically diverse and have many valuable alleles (Lee et al., Reference Lee, Yu, Hwang, Blake, So and Lee2008).
Recent advances in sequencing technologies has also highlighted the unique genomic structures of both cultivated and wild soybeans and has provided an opportunity to use G. soja to broaden the genetic base of cultivated soybean (Kim et al., Reference Kim, Diers, Hyten, Roufmian, Shannon and Nelson2012; Joshi et al., Reference Joshi, Valliyodan, Wu, Lee, Xu and Nguyen2013). Furthermore, assessing genomic differences for important traits will provide insights into the process of speciation and domestication and will deepen our understanding of the origin of genes involved in complex traits (Kim et al., Reference Kim, Lee, Van, Kim, Jeong and Choi2010; Joshi et al., Reference Joshi, Valliyodan, Wu, Lee, Xu and Nguyen2013; Li et al., Reference Li, Zhao, Ma, Li, Yan, Li, Qi, Guo, Zhang, He, Chang, Liang, Guo, Ye, Wang, Tao, Guan, Wang, Liu, Jin, Zhang, Liu, Zhang, Chen, Wang, Nielson, Li, Chen, Li, Reif, Purugganan, Wang, Zhang, Wang and Qiu2013; Zhao et al., Reference Zhao, Zheng, He, Wu, Pan and Lam2015; Zhou et al., Reference Zhou, Wang, Jian, Geng, Wen, Song, Wu, Li, Liu, Dunwell, Zhang, Feng, Niu, Zhang, Ren and Zhang2015; Han et al., Reference Han, Zhao, Liu, Li, Lightfoot, Yang, Zhao, Zhou, Wang, Huang, Zhang, Qiu, Zheng and Li2016).
Isolation and characterization of seed storage proteins from wild soybeans are essential to physiological studies of protein accumulation during development and their utilization during germination. In addition, sucrose contributes to good taste and desirable texture of soyfoods, and the sucrose content determines the sweetness of raw vegetable soybean seed (Kumar et al., Reference Kumar, Rani, Goyal, Pratap, Billore and Chauhan2011). Furthermore, sucrose content is higher when soybeans are grown at cooler locations, whereas the stachyose content is genotype-dependent across different growing environments (Kumar et al., Reference Kumar, Rani, Goyal, Dixit, Manjaya, Dev and Swamy2010). Another factor that has made breeding for the sucrose content in soybean seed difficult is the cost for quantifying this disaccharide (Maughan et al., Reference Maughan, Maroof and Buss2000). Conventional analytical chemistry techniques for seed constituents are expensive, time consuming and destructive. Fourier transform near-infrared reflectance spectroscopy (NIRS) technology was introduced over the last few decades for the wide-scale, inexpensive chemical analysis of food and crop seed composition (Wilcox and Cavins, Reference Wilcox and Cavins1995; Buning and Diller, Reference Buning and Diller2000).
Here, we present the distribution and characteristics of three major classes of seed components, protein, lipid and sucrose, based on NIRS measurements of 334 wild soybean accessions from regions representing a wide range of climatic conditions in Japan.
Materials and methods
Plant materials and collection sites
A total of 334 Japanese wild soybean (G. soja) accessions were used in this study. All accessions were collected across several climatic zones from Niikappu, Hokkaido (42°25′06″N) to the southernmost site, Minamikyushu, Kagoshima (31°16′11″N). The collection sites of the wild accessions covered a broad geographic range in Japan and are listed in Table S1.
All accessions were classified into the macro (Roman numeral) and meso (Arabic number) subdivisions of climatic regions as described by Yoshino (Reference Yoshino1980). For simplicity, the collection sites were classified into geographic areas according to the climatic classification in Japan (Figure 1). These accessions were obtained from ‘Legumebase’, an online database (http://www.legumebase.brc.miyazaki-u.ac.jp/) supported by the National BioResource Project in Japan. For the comparison, a G. max cultivar, ‘Fukuyutaka’, was evaluated for seed protein, lipid and sucrose contents using identical methods as those used for the wild accessions.
Fig. 1. The collection sites of Glycine soja accessions used in this study. Axes indicate latitude and longitude. Macro (Roman numerals) and meso (Arabic numbers) scale subdivisions indicate the climatic regions of Japan according to Yoshino (Reference Yoshino1980).
Quantitative analysis of protein content
Seed protein in the wild accessions was measured quantitatively using the Biuret method. Seed coats were removed from 20 to 30 seeds per accession, the naked seeds were ground using a TissueLyser II (QIAGEN, Hilden, Germany), and finally the particles were filtered through a 0.5 mm sieve. Soybean powder was dried in a 100°C dryer for 1 h and left to cool in a desiccator at room temperature for 1 h. Soybean powder (50 mg) was placed in a flask to which 0.5 ml of chloroform (CHC13), Biuret reagent (10 mM KOH, 5 mM KNaC4H4O6 4H2O) and 20 ml of 1.6 mM Cu2SO4 5H2O were added. The flask was shaken for 15 min and held for 1 h without shaking. The flask contents were transferred to a 50 ml tube and centrifuged at 3300 × g for 15 min. The absorbance of the supernatant liquid was measured at 550 nm using a Nano Drop 1000 spectrophotometer (Thermo Fisher Scientific, Inc., MA, USA).
Quantitative analysis of lipid content
Seed lipids were quantitatively measured from the wild accessions using a CHC13 extraction method (Southgate, Reference Southgate1971) with a few modifications. Soybean seeds (100 mg) were ground into a fine powder and transferred into a 2-ml Eppendorf tube containing an iron bead. After adding 1.5 ml of hexane, the mixture was extracted by shaking in a multi-beads shocker (Yasui Kikai Corporation, Osaka, Japan) at 800 × g for 60 s, followed by centrifugation at 5100 × g for 15 min. The aqueous phase was collected and transferred to a 13 × 100 mm glass test tube, and the remaining pellet was re-suspended in 1 ml of hexane. The above extraction steps were repeated three times. For purification, the lipid-containing solution was washed with 3 ml of H2O, mixed gently by inversion, and centrifuged at 800 × g for 15 min. The upper phase was removed, transferred to a clean glass test tube using a Pasteur pipette, and filtered through a 0.2 µm pore size syringe filter. Next, the hexane solution was evaporated at 70°C in a Dry Thermo Unit TAH-2 G (TAITEC Corporation, Saitama, Japan) for 4–6 h, followed by drying in a SpeedVac-SC100 centrifugal evaporator (SAVANT INSTRUMENTS, INC, NY, USA). Finally, the lipid was dissolved in 0.5 ml of CHC13 and transferred to a weighed fresh glass test tube by means of Pasteur pipette. This final step was repeated three times to thoroughly rinse the inner part of the test tubes. After evaporating the CHC13 at 70°C in a Dry Thermo Unit for 2 h, the test tube was weighed again. The difference in weight was considered to be the lipid weight, and the lipid content was expressed as a percentage of the 100 mg ground-seed weight.
Lipid content (%) = (W 1–W 0) × 100
W 1, weight of the tube containing the extracted lipid; and W 0, weight of the empty tube.
Quantitative analysis of sucrose content
Quantitative measurements of seed sucrose content in the wild accessions were obtained using a high-performance liquid chromatograph (HPLC) (SHIMADZU Corporation, Kyoto, Japan) method after a preliminary fat extraction. Seed coat were removed and ground using a TissueLyser II (QIAGEN, Hilden, Germany), and finally the particles were filtered through a 0.5 mm sieve. Soybean powder (0.15 mg) was mixed with 1.5 ml of distilled water in a 2-ml Eppendorf tube. The mixture was shaken horizontally at 150 rpm for 15 min using a Uni Thermo shaker NTS-1300 (TOKYO RIKAKIKAI CO., LTD., Tokyo, Japan) at room temperature, and centrifuged at 15,300 × g) for 10 min at room temperature. The supernatant liquid (0.5 ml) was transferred to a 1.5 ml tube, 0.5 ml of 95% acetonitrile was added at room temperature, and the tube was centrifuged again. The supernatant liquid (0.2 ml) was transferred into a new 1.5 ml test tube and dried at 95°C in a dry bath incubator for 45 min. Finally, the residue was dissolved in 0.2 ml of 65% acetonitrile and filtered through a 0.2 µm syringe filter prior to HPLC analysis.
Sucrose was identified and quantified using a TSK-GEL amide column (80 × 5μm, column temperature: 50°C) (TOSOH corporation, Tokyo, Japan) in a CBM-20A HPLC system equipped with a LC10AD pump, a degasser DGU-14A, an auto sampler SIL-10AD, a system controller CBM-20A, a column thermostat CTO-14A and a refractive index detector RID-10A (SHIMADZU corporation, Kyoto, Japan). Sucrose was eluted with 75% (v/v) acetonitrile and 25% (v/v) distilled water at a flow rate of 0.5 ml/min under isocratic conditions.
Seed composition analyses for protein, lipid and sucrose using NIRS
Fourier NIRS (NIRFlex N-500, BUCHI Labortechnik AG, Flawil, Switzerland) was used to develop calibration equations for simultaneously measuring the protein, lipid and sucrose contents of seeds. Seed coats of 334 wild accessions were removed, approximately 50–100 naked seeds per accession were pulverized using a TissueLyser II (QIAGEN, Hilden, Germany), and finally the particles were filtered through a 0.5 mm sieve. The interval for the spectrometric analyses was 4 cm−1 within a wavenumber range of 4,000–10,000 cm−1; the near-infrared range is 800–2,500 nm (4,000–12,500 cm−1). The preconditioning process, including smoothing, differentiation and normalization of spectrometric analyses, was carried out to revise environmental factors and spectrometer confusion using NIRCal (BUCHI Labortechnik AG, Flawil, Switzerland). The linear combination of variables for absorbance changes in near-infrared wavelength range was translated to explanatory variables, and a partial least-squares (PLS) regression was conducted using the chemical analyses values as target variables.
The equation for the PLS regression algorithm was
$$y_i = \mathop \sum \limits_{k = 1}^r c_k \cdot t_{ik} + e_i, $$
where y i is the target variable (chemical analysis value) of the sample i; c k is a regression coefficient for wavelength k; t ik is an explanatory variable for sample i and wavelength k; and e i is a potential variable for the residual error of sample i. Moreover, the accuracy of calibration was estimated by the Evaluation index (EI) testing method (Mizuno et al., Reference Mizuno, Ishiguri, Kondo and Kato1988).
$${\rm EI} = \displaystyle{{2 \times {\rm SDP}} \over {{\rm Range}}} \times 100{\rm \;} ({\rm \%} ),$$
where SDP means the standard deviation of the prediction error; and Range means the range of prediction for sample values.
Results
Set of calibration curves and accuracy tests for the NIRS results
We measured the protein, lipid and sucrose contents of seeds from wild and cultivated soybeans using NIRS, and designed a calibration curve set from the results obtained by chemical analysis. A total of 50 accessions were examined for the calibration set; the correlation coefficients (r 2) and standard error of calibration (SEC) values for the protein, lipid and sucrose contents were 0.917 and 1.999, 0.899 and 1.473, 0.758 and 0.771, respectively (Table 1). In order to estimate the accuracy of the calibration curve, an EI test was conducted using this calibration curve set. From the NIRS analyses, the range, r 2 and SDP values for protein, lipid and sucrose contents are presented in Table 1. In accordance with the accuracy testing method, the EI values for protein, lipid and sucrose were calculated to be 20.0, 21.6 and 22.9%, respectively. All EI values for seed composition as measured by NIRS analysis were within a 12.5–24.9% range, and the criterion based on EI values was classified as a rank B, which has high accuracy and good practical usability (Mizuno et al., Reference Mizuno, Ishiguri, Kondo and Kato1988). These results indicate that NIRS can be used to measure these three chemical components instead of using chemical analysis methods.
Table 1. Calibration set and accuracy test of NIRS results
a Rank A = under 12.4%, B = 12.5–24.9%, C = 25.0–37.4%, D = 37.5–49.9%, and E = upper 50.0% of EI value (Mizuno et al., Reference Mizuno, Ishiguri, Kondo and Kato1988).
Evaluation and distribution of protein, sucrose and lipid contents in wild soybean seeds
Three hundred and thirty-four accessions of wild soybean collected in Japan were evaluated for their seed protein, lipid and sucrose contents by NIRS-measured frequencies relative to the respective calibration equations. Figure 2 shows the distributions and ranges of the three constituents in the same sample set. The distribution of protein content ranged from 48.6% (B03039) to 57.0% (B02120), and 196 accessions (~59%) of all wild soybean accessions had protein contents of 54%, the greatest frequency (Figure 2A). One, 4, 53 and 80 wild accessions had protein contents of approximately 48, 50, 52 and 56%, respectively. As the protein content of the G. max. ‘Fukuyutaka’ was 44.6%, the protein content of all wild accessions was higher than that of the control, suggesting that the protein content of G. soja seeds is greater than that of G. max.
Fig. 2. Distribution of protein, lipid and sucrose contents of Glycine soja accessions collected in Japan (G.m = Glycine max ‘Fukuyutaka’).
The lipid content of the wild accessions varied from 9.0% (B02156) to 14.3% (B03039), and 158 accessions had a lipid content of approximately 11%, the highest frequency for this seed component (Figure 2B). Four, 54, 90, 23 and 5 accessions had lipid contents of approximately 9, 10, 12, 13 and 14%, respectively. Furthermore, the lipid content of ‘Fukuyutaka’ was 20.1%, suggesting that the lipid content of G. soja tends to be lower compared with G. max.
Among the wild accessions, the seed sucrose content varied from 1.24% (B02237) to 3.53% (B02258), and 158 accessions had sucrose contents of approximately 2.5%, the highest frequency for this seed constituent (Figure 2C). One, 14, 105, 51 and 5 accessions had approximately sucrose contents of 1.0, 1.5, 2.0, 3.0 and 3.5%, respectively. The sucrose content of Fukuyutaka with 6.89%, suggests that G. soja tends to have a lower sucrose content than G. max, a result similar to the pattern for lipid content.
Correlations between the protein, sucrose and lipid contents in wild soybean seeds
Linear regressions between protein, lipid and sucrose contents were calculated using 334 Japanese wild soybean accessions, and Figure 3 shows the results and correlation coefficients. High negative correlations were observed between lipid and sucrose contents and protein content, and the r 2 values were 0.5548 and 0.4922, respectively (Figure 3A and C). Moreover, the correlation between the sucrose and lipid contents was highly positive with a 0.6427 r 2 value (Figure 3B). In contrast, the ratio of the lipid and sucrose contents as a function of the protein content in G. soja accessions was lower than that of ‘Fukuyutaka’ (Figure 3A and C). Similarly, the ratio of the sucrose content as a function of the lipid content in G. soja accessions was also lower than that of ‘Fukuyutaka’ (Figure 3B).
Fig. 3. Correlation between protein, lipid and sucrose contents of Glycine soja accessions collected in Japan. Solid circles: G. soja, open circles: Glycine max ‘Fukuyutaka’.
Comparison of seed compositions of plants collected from various climatic regions in Japan
Seed protein content means for G. soja plants collected from regions II-1, II-2, III-1, III-2, III-3, III-4, III-5, IV-1, IV-2, IV-3 and V-1 were 54.6, 55.2, 53.9, 54.1, 53.7, 54.3, 54.2, 53.7, 54.0, 54.1 and 54.1%, respectively (Figure 4A). The total seed protein content of G. soja collected from regions with high annual average temperatures was high, especially the protein content for accessions from the II-1 region were significantly higher than those from region III-3. The sucrose content means in the same respective climatic regions described above were 2.19, 2.02, 2.36, 2.44, 2.42, 2.43, 2.48, 2.42, 2.47, 2.43 and 2.24% (Figure 4C). The seed sucrose content of accessions collected in region II-1 was significantly lower than those from regions III-2 and IV-3. Although it was not significant, the seed lipid content of accessions from region II-1 was lower than those from the other regions (Figure 4B). The G. soja accessions collected in region II had higher protein contents and lower sucrose and lipid contents than the other regions.
Fig. 4. Comparison of protein, lipid and sucrose contents of Glycine soja collected from distinct climatic divisions in Japan. The different letters on the bars show significance at P < 0.05 by the Steel-Dwass test.
Discussion
NIRS is time-efficient analytical technology for screening a large number of samples for different chemical constituents and their concentrations (Font and de Haro-Bailon, Reference Font and de Haro-Bailon2006). Therefore, the NIRS technique has been utilized to determine the protein and oil content of progeny during plant breeding and has been used to measure concentrations of fatty acids, anthocyanins, 11S to 7S protein fractions, isoflavones, lecithin, oligosaccharides and amino acids in soybean (Lee et al., Reference Lee, Shannon and Choung2011). The availability of NIRS calibrations for screening breeding materials is especially desirable because a single NIRS spectrum can be collected to measure seed protein and oil content. Furthermore, the NIRS-based method does not require additional sample preparation as compared with chromatographic or enzymatic methods for sucrose analysis (Choung, Reference Choung2010; Sato et al., Reference Sato, Zahlner, Berghofer, Losak and Vollmann2012).
In this study, we investigated the protein, lipid and sucrose contents of wild soybean (G. soja) accessions collected in Japan using NIRS. The collection sites for the 334 accessions were selected to represent populations present in diverse regions throughout Japan. We investigated the abundance of the three most important seed components in these accessions to determine the degree of diversity present in seeds collected from different climatic areas.
Although there remains a strong economic incentive to develop cultivars with high protein and oil contents, while maintaining a competitive yield, little progress towards these goals has been made. Effective breeding for seed composition requires accurate, inexpensive and reliable techniques for measuring seed constituents. Certain areas of breeding have benefited from methods to analyse a single soybean seed or even parts of seeds.
Numerous investigations of the regional variations in seed composition in soybean cultivars have been reported (Hurburgh et al., Reference Hurburgh, Brumm, Guinn and Hartwig1990; Hurburgh, Reference Hurburgh1994; Piper and Boote, Reference Piper and Boote1999; Yaklich et al., Reference Yaklich, Vinyard, Camp and Douglass2002; Brumm and Hurburgh, Reference Brumm and Hurburgh2006; Dardanelli et al., Reference Dardanelli, Balzarini, Martinez, Cuniberti, Resnik, Ramunda, Herrero and Baigorri2006). Among these investigations, the protein and lipid contents of soybean cultivars in America were reported to be in the approximate ranges of 33–48 and 17–21%, respectively. According to a study on seed components of wild soybean in China (Dong et al., Reference Dong, Zhuang, Zhao, Sun and He2001), 5929 accessions had an average protein content of 45.42%, with significant differences seen among accessions collected at different latitude zones. There were also extreme cases of protein content among the Chinese accessions: the highest protein content was 55.7% and the lowest was 29.0%. The average lipid content of the wild accessions in China was 10.6% with accessions varying in lipid content from 4.4 to 20.2%.
In our study, the seed protein content varied from 48.6 to 57.0% and had an average of approximately 54.0%; the lipid content varied from 9.0 to 14.3% and had an average value of 11.4%. The results of our investigation indicate higher protein and lower lipid contents in the wild accessions than those reported in other studies. This result might be due to the method used to measure the seed components; however, the seed composition characteristics of wild soybean accessions collected in Japan have not previously been investigated.
Many studies have discussed the inverse relationship between seed protein and oil contents (Brim and Burton, Reference Brim and Burton1979; Carver et al., Reference Carver, Burton, Carter and Wilson1986; Dornbos and Mullen, Reference Dornbos and Mullen1992; Bellaloui et al., Reference Bellaloui, Smith, Ray and Gillen2009), but little is known about the relationships among protein, oil and sugars, even though sugars can contribute to more than 10% of the total seed dry weight. A previous study indicates that the total sugar and oil concentrations in soybean seeds are positively associated, and each is negatively correlated with protein concentration (Openshaw and Hadley, Reference Openshaw and Hadley1978). Breeding efforts to modify seed composition have concentrated on changing only the oil and protein percentages while ignoring the sugar concentration. Sucrose was the predominant soluble sugar, accounting for an average of 55.16% (33–90%) of the total sugar. The average sucrose content was 5.35% among all G. max accessions screened, with more than threefold differences between the extremes, ranging from 2.98 to 10.55% (Hou et al., Reference Hou, Chen, Shi, Zhang and Wang2009). Our results indicate that the sucrose content of Japanese wild soybean varied from 1.24 to 3.53%.
In this paper, we classified the collection sites based on the climatic regions with reference to the standard scale of climates in Japan (Yoshino, Reference Yoshino1980). The boundaries are established as follows. The boundary between regions II-1 and II-2 approximately coincides with the 180°C warmth index line, or the 20°C isotherm for annual air temperature. The boundary between regions II and III coincides in most parts with the 0°C isothermal line for the mean daily minimum temperature in January, which is an indicator for frequent frost occurrence. This boundary roughly corresponds to the 16°C isotherm of the annual mean temperature on the Pacific coast of Honshu, Shikoku and Kyushu. The boundary between regions III and IV, which is an important boundary in Japan, reflects the divide between the Pacific climate and the Japan Sea climate. The boundary is drawn in accordance with the 50 cm isoline for mean maximum snow depth that plays a defining role as a delimiting factor in the distribution of vegetation. The boundary between regions IV and V is drawn by taking into account the regions where the monthly mean temperature is 0°C for 4 months or more (Yoshino, Reference Yoshino1980).
Although no wild accessions from region I were evaluated in our study, the seed composition of accessions from region II showed interesting results, especially as related to protein and sucrose contents. The average value for protein content in accessions collected in region II was higher than the other regions, and the average value for the sucrose content was low. These results indicate that the wild soybean accessions from the region with the highest annual temperature in Japan were characterized as having high seed protein levels and low lipid and sucrose levels.
The wild soybean is the direct progenitor of the soybean cultivar (Hymowitz, Reference Hymowitz1970) is distributed in Russia, Korea, Japan and China. Because of the geography of Japan (north (46°N, 146°E) and south (24°N, 122°E)) variation in photoperiod and temperature are great; thus, ecotype divergence may also vary according to the regional adaptation (Kaga et al., Reference Kaga, Shimizu, Watanabe, Tsubokura, Katayose, Harada, Vaughan and Tomooka2012). Information of geographic distribution and genetic diversity of wild and cultivated soybeans in Japan, including other countries is important especially in breeding programmes and conservation of soybeans. Previous studies have reported low genetic exchanges between germplasm from northern, central and southern regions in Japan (Zhou et al., Reference Zhou, Carter, Cui, Miyazaki and Burton2002). Evaluation of the genetic structure of wild soybean in Japan based on microsatellite markers and found that soybean varieties accounting for 95% of the cultivated area in Japan had much lesser genetic variation as compared with that of the wild soybeans (Kuroda et al., Reference Kuroda, Kaga, Tomooka and Vaughan2006). This trend is supported by other previous studies for wild and cultivated soybeans using germplasm from other countries (Maughan et al., Reference Maughan, Saghai and Buss1995; Xu et al., Reference Xu, Abe, Gai and Shimamoto2002) using microsatellite variation. Moreover, geographical structure of genetic variation in Japan indicated that wild soybeans in northern Japan are distinct from southern Japan, while wild soybeans in central Japan had variations from both regions (Kuroda et al., Reference Kuroda, Kaga, Tomooka and Vaughan2006). On the other hand, Zhou et al. (Reference Zhou, Carter, Cui, Miyazaki and Burton2002) reported distinct genetic base of Japanese improved varieties when compared with that of China, USA, Canada. Comparison between Chinese soybean ecotypes from the three large ecoregions with improved Japanese varieties found that Japanese varieties had a lower genetic diversity than Chinese ecotypes but distinct alleles from them and suggested the potential to broaden the genetic base (Guan et al., Reference Guan, Chang, Li, Wang, Liu and Qiu2010). With the availability of high-throughput genotyping platforms, SNPs markers based could estimate more precise estimation of genetic relationships. Recently, SNP markers were used to characterize the genetic variation and population structure of 1603 soybean accessions: 832 Japanese landraces, 109 old and 57 recent Japanese varieties, 341 landrace from 16 Asian countries and 264 wild soybean accessions, and it was revealed that the gene diversity of soybean in Japan was slightly lower than that of exotic soybean germplasm; however, there was a clear genetic differentiation among Japanese cultivated soybeans, exotic cultivated soybeans and wild soybeans as indicated by population differentiation and clustering analysis (Kaga et al., Reference Kaga, Shimizu, Watanabe, Tsubokura, Katayose, Harada, Vaughan and Tomooka2012).
In this study, our results provide diverse and wide-ranged seed chemical composition information of wild soybean resources according to the climatic region for improving of new soybean varieties in global environmental changes. And the findings from this investigation will be useful for modifying soybean seed composition in a breeding programme using wild soybean accessions. A major objective of this study was to correlate the effects of environmental conditions on wild seed composition in order to predict regions and environmental conditions that consistently produce high-quality soybeans. As a direct result of this project, it will be feasible to use the wild soybean accessions in the future development and selection of new soybean varieties with desired traits.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262116000472
Acknowledgement
This work was supported by the National Bio-Resource Project of the Japanese Ministry of Education, Culture, Sports, Science and Technology (Mext), Japan.
